CopySpec accelerates LLM inference by copying repeated token sequences from context and verifying them with the target model, achieving up to 3.08x speedups without altering outputs.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CL 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
CopySpec: Accelerating LLMs with Speculative Copy-and-Paste Without Compromising Quality
CopySpec accelerates LLM inference by copying repeated token sequences from context and verifying them with the target model, achieving up to 3.08x speedups without altering outputs.